In [ ]:
# ==== Point 1: Plot 8–10 sample images per class ====
import os, glob, random, math
import cv2
import matplotlib.pyplot as plt
# Your actual folders (post-unzip)
TRAIN_DIR = "/content/data_structures/train/Stuctures_Dataset"
TEST_DIR = "/content/data_structures/test/Dataset_test/Dataset_test_original_1478"
# Config
SAMPLES_PER_CLASS = 10 # set to 8–10 as required
IMG_SIZE = (224, 224) # display size (not training preprocessing)
RANDOM_SEED = 42
random.seed(RANDOM_SEED)
# Utility: list class folders
def list_classes(root):
return sorted([d for d in os.listdir(root) if os.path.isdir(os.path.join(root, d))])
# Utility: gather image paths for a class (common extensions)
IMG_EXT = (".jpg", ".jpeg", ".png", ".bmp", ".tif", ".tiff", ".webp", ".JPG", ".PNG")
def get_image_paths(folder):
return [p for p in glob.glob(os.path.join(folder, "*")) if p.endswith(IMG_EXT)]
# Utility: read image with OpenCV (BGR->RGB) + optional resize
def read_rgb(path, target_size=None):
img = cv2.imread(path)
if img is None:
raise ValueError(f"Failed to read: {path}")
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
if target_size is not None:
img = cv2.resize(img, target_size, interpolation=cv2.INTER_AREA)
return img
# Helper to plot a grid of images for a single class
def plot_class_samples(class_name, class_dir, samples=10, img_size=(224,224)):
paths = get_image_paths(class_dir)
if not paths:
print(f"[WARN] No images found for class '{class_name}' in {class_dir}")
return
k = min(samples, len(paths))
sel = random.sample(paths, k)
cols = 5 if k > 5 else k
rows = math.ceil(k / cols)
plt.figure(figsize=(3*cols, 3*rows))
plt.suptitle(f"Class: {class_name} | showing {k} sample(s)", fontsize=14)
for i, p in enumerate(sel, 1):
img = read_rgb(p, target_size=img_size)
plt.subplot(rows, cols, i)
plt.imshow(img)
plt.title(os.path.basename(p)[:30], fontsize=8)
plt.axis("off")
plt.tight_layout(rect=[0, 0, 1, 0.95])
plt.show()
# --- Sanity: list classes and counts (train + test) ---
def count_by_class(root):
classes = list_classes(root)
return {c: len(get_image_paths(os.path.join(root, c))) for c in classes}
print("Train classes & counts:")
train_classes = list_classes(TRAIN_DIR)
print({c: count_by_class(TRAIN_DIR)[c] for c in train_classes})
print("\nTest classes & counts:")
test_classes = list_classes(TEST_DIR)
print({c: count_by_class(TEST_DIR)[c] for c in test_classes})
# --- Plot 8–10 samples per class for TRAIN set ---
for cls in train_classes:
plot_class_samples(cls, os.path.join(TRAIN_DIR, cls),
samples=SAMPLES_PER_CLASS, img_size=IMG_SIZE)
#visualize TEST set samples per class
for cls in test_classes:
plot_class_samples(cls, os.path.join(TEST_DIR, cls),
samples=SAMPLES_PER_CLASS, img_size=IMG_SIZE)
Train classes & counts:
{'altar': 829, 'apse': 514, 'bell_tower': 1059, 'column': 1919, 'dome(inner)': 616, 'dome(outer)': 1177, 'flying_buttress': 407, 'gargoyle': 1571, 'stained_glass': 1033, 'vault': 1110}
Test classes & counts:
{'altar': 140, 'apse': 57, 'bell_tower': 170, 'column': 210, 'dome(inner)': 86, 'dome(outer)': 168, 'flying_buttress': 78, 'gargoyle': 240, 'stained_glass': 163, 'vault': 164}
